Minutes for LISP at IETF-95
Locator/ID Separation Protocol
||Minutes for LISP at IETF-95
Session 1/1 (120 Minutes)
Thursday, April 7, 2016
1730-1930, Afternoon Session III, 120 Minutes
Room Atlantico B
- Blue Sheets
- Agenda Bashing
- Status reports for WG drafts
o WG Documents Update
- LCAF Draft Update - draft-ietf-lisp-lcaf-12
• Dino Farinacci asks if the document is ready.
• Fabio Maino says the document is ready.
• Chair asking the audience: there is no objection.
- LISP-crypto - draft-ietf-lisp-crypto-03
• new changes reflect discussion with crypto specialist and choice for
high performance. • implementation show that in software it still
offers good performances. • Dino Farinacci asks for other
implementations to test. • Chair asking the audience if document is
ready: no objection.
- LISP SF Multicast - draft-ietf-lisp-signal-free-multicast-00
• Robert Raszuk proposes to use LISP to implement multicast and let
o Non WG Items
- Open Overlay Router
Fabio Maino (on behalf of A. Cabellos (excused))
• Open Overlay Router is a re-focus of the LISPmob project from
mobility to overlay router. • Joel Halpern asks if it is an alternative
or a complement to large open source projects. • Fabio Maino explains
that there is current effort to integrate fd.io and Open Overlay Router.
- ComCast IPv6 SR
• presentation cancelled (excused)
- GPE-VPN - draft-maino-gpe-vpn-00.txt
• Discussion from the audience to clarity the next protocol field.
• Joel Halpern reminds that the number of services will remain modest.
- Unified L2 and L3 overlays using the LISP control plane
• Padma Pillay-Esnault: how long is a negative mapping kept? How fast
to detect? • Victor Moreno: we keep until a timeout or SMR trigger •
Padma Pillay-Esnault: might be an issue when mobility. • Dino
Farinacci: when a device move, it does not leave it so when it
re-attachs the network it will trigger a change. If the node is gone,
then it is timeout. • Victor Moreno asks the group to review the draft
and if the document is a good reference for the group. • Chairs: it is
to early to talk about adoption • Dino Farinacci: it is important to
keep in mind that everything moves (containers, VMs, … ). The big
difference with lisp mn is that RLOCs and EIDs are not collocated in
this L2/L3 overlay
- IANA Registry for LISP Packet Type Allocations -
L. Iannone (on behalf of M. Boucadair)
• Fabio Maino: having an experimental type makes sense.
• Dino Farinacci proposes to encode the type in JSON.
- LISP Mapping Service Discovery at Large -
S. Secci (on behalf of M. Boucadair)
• Dino Farinacci: it is not a good idea to put overlay information in
the underlay. • Dino Farinacci: to avoid import/export it would be
better to use something like DDT and use instance-id to distinguish the
different tenants. • Stefano Secci: the objective is to “automatically”
connect independent mapping systems. • Robert Raszuk says that the
problem is already solved with route servers. • Padma Pillay-Esnault
says that using extended communities is dangerous as it should be per
prefix and there is a risk of aggregation.